Problems, Solutions, and Optimization for Modern Clouds

Professor Lau, Francis Chi Moon   (Project Coordinator (PC))

Professor Lee Patrick Pak-ching   (Co-principal investigator)

Professor Chen Minghua   (Co-principal investigator)

Dr Xu Hong, Henry   (Co-principal investigator)

Dr Wang Jianping   (Co-principal investigator)

Professor Li Bo   (Co-principal investigator)

Professor Wu Chuan   (Co-principal investigator)

Professor Chen Kai   (Co-principal investigator)

40

2016-03-01

5102143

Problems, Solutions, and Optimization for Modern Clouds

Cloud computing, Datacenter, Optimization, Resource allocation, Virtualization

Network,Computer Science Fundamentals

Engineering (E)

C7036-15G

Collaborative Research Fund (CRF) - Group Research Project

2015

Completed

1. To model the relationships existing among performance, workloads, resources of composite cloud applications and to design cost-efficient resource scaling mechanisms based on the models in multi-tenant settings. 2. To devise online algorithms and pricing mechanisms for dynamic resource provisioning and assembly, which will lead to optimal resource utilization and maximum social welfare over the long run of the cloud system. 3. To design an efficient mechanism to handle tenants' bandwidth demands that can provide bandwidth guarantees, satisfy work-conservation, and offer low latency to short flows. 4. To study the effects of the scheduling delays introduced by the hypervisor on the functioning of the VM and the guest OS and to devise methods to help those functions that are delay-sensitive. 5. To develop an optimization framework, fast algorithms, and a practical failover protocol to improve efficiency and robustness of bandwidth scheduling in inter-datacenter WAN. 6. To design dynamic provisioning schemes that, via scheduling of servers and power generators and renewable energy sources, can effectively reduce datacenter energy costs. 7. To build a cloud management system to enable deployment and evaluation of our algorithms and designs in real world cloud platforms as well as large cloud testbeds.